Like many people these days, I have been playing and working with Artificial Intelligence (AI) and have been absolutely amazed at what it can do, particularly in connection with what I spend much of my days doing – thinking, researching, and writing. Earlier versions of AI with which I tried working turned out texts that were, what I would refer to as, competent, but relatively soulless and lacking personality. However, newer versions can turn out works that have personality, charm, and occasionally humor. I can even instruct the AI to produce a piece in my own writing style, and it has shown itself to be a reasonably good mimic.
I look at my professional relationship with AI as a collaboration between the AI and me. For example, when deciding upon what to blog on, I will now frequently request the AI to prepare a draft according to certain specifications – a request that the AI is more than happy to oblige. It will even churn out different versions until it has produced a draft that I feel I can work with. I then use that draft as a starting point and edit in whatever of my own knowledge, research, and thinking the AI has not generated on its own. Even though the AI has gotten better at matching my writing style, I still edit the text so that it reads authentically in my voice. The result is a version that combines the best of what I and the AI have to offer and that I can then post. This is the sort of beneficial and creative collaboration with AI that Ethan Mollick touts in his new book, Co-Intelligence, which I highly commend to any of you who have an interest in AI, its capabilities (and weaknesses), and how to enhance productivity by working with it.
AI is now advanced enough that it can turn out serviceable work product that in many cases does not require further editing and input in order to be considered adequate for the applicable purposes. However, I am very much concerned that this aspect of AI is going to make a lot of people lazy. While the diligent among us will not just rely on the AI’s output as the final product, I have no doubt that many people will present work produced entirely, or almost entirely, by the AI as their own and turn that product in or cut-and-paste it without any further research or editing. Simply put, my fear is that AI will enable laziness among users, be they students, workers, or artists. I am concerned that this will encourage an indolence and slothfulness among humans to the detriment of the intellectual and cultural condition and progress of our species. (This observation applies not just to written text, but, mutatis mutandis, to any AI work product, whether it be an article, code, or work of art.) Just as auto-correct has been shown to have had a detrimental effect on spelling, AI runs the risk of depleting our writing and other expressive aptitude.
I would note that merely cutting-and-pasting AI text is not only lazy, but unwise, and that the user does so at their own risk. First, as asserted above, work that the AI user has had a hand in producing through collaboration will be superior to what the AI has produced on its own. Second, it is well-known that, for all its technological brilliance, AI still screws up occasionally. In this regard, Mollick notes that AI will occasionally make things up (what is referred to as “hallucinating”). Accordingly, the user would be wise to check what the AI produces carefully, lest they turn in mendacious and defective work. In addition, there have been cases where text produced by AI has included racist, sexist, and other offensive elements. While some of this has occurred in cases where those communicating with the AI have tried to manipulate or trick the AI into generating offensive content, AI can be unpredictable and turning in an unexpectedly offensive AI text could very well result in a failing grade, firing, or other negative ramifications for the user. Imagine how damaging to one’s career it could be if a user handed in a project to a boss which included misogynistic or scatological passages. I believe that few bosses would see the humor in it.
Even with these risks, however, many people will undoubtedly just use what the AI generates, figuring that hallucinations and offensive language happen too infrequently to worry about. And, as AI improves, these problems will occur less and less, giving users even greater confidence that turning in work produced by an AI is not a risky venture. Accordingly, my concerns about AI’s potential contribution to human laziness will become increasingly valid.
How far will this threat to the human intellect go? When I reflect on this question, my thoughts turn to the Eloi of H.G. Wells’s The Time Machine. For those of you who are not familiar with that story, an inventor travels far into the future using a time machine he has built. At that point, the human species has cleaved into two separate species. One group is the troglodytic, brutish, and subterranean-dwelling Morlocks, whom the time traveller believes are the evolutionary descendants of the working class. The other group is the Eloi, who are believed to be descended from the upper classes. Due to thousands of years of having everything done for them, the Eloi have devolved into weak, helpless, and feeble-minded beings. (While the Eloi are, admittedly, a happy people, that is a result of their ignorance and lack of intellectual capabilities.) The Morlocks provide for the material needs of the Eloi and, in turn, feed on them.
What the brilliant allegory of the book is saying is that people who fail to invest their human faculties and who merely exploit others for their needs are destined to, over time, degenerate, bringing civilization down with them. Seen through this lens, the lazy who merely pass on the work of AI without making a thoughtful contribution to the product may end up as pathetic as the Eloi. I appeal to all of those out there who are making use of AI not to merely pass on what AI generates, but instead to regard the AI as a collaborator in your work and to fully engage yourselves in producing the final work product. AI should enhance and help build our intellectual capabilities, not substitute for them. Otherwise, I fear that our evolutionary descendants will end up as unintelligent and decadent as the Eloi.
